Right on, Jake.
In addition, one has to remember that the 'drive specifications' quoted by all the drive makers are a little like the 'gasoline mileage' statements made by the automobile manufacturers. In fact, the 'drive specifications' should probably include the statement 'your mileage may differ' because the ultimate performance of a system does not hinge solely on the 'seek time' or 'track to track time' or 'burst data transfer time' or any of the other popular specification points quoted.... A system's performance will depend on the operating system, the file allocation system (FAT, HPTS, FAT32, etc), the caching algorithms, the fragmentation level of the data on the drive, etc. AND probably most important of all is the way the actual subject program is using the DASD and the data contained on it.
Why should I care if the 'seek time'specification , which is an artificial measure of the average time needed to traverse 1/3 the tracks on a platter, is one millisecond different than another drives?
Important questions relative to 'avg seek time' are:
Do the drives have the same number of tracks per platter?
How much data is contained on each track?
If a drive has more data per track, then when you traverse 1/3 of the tracks you would have accessed over more data on that drive. Does accessing over more data per operation make a drive better?
If a drive has more tracks per platter, then traversing 1/3 of the tracks access over more data than another drive. Ditto?
This specification business is partly a game of words and how to present them to give your product what looks like the best light. I know 'cause I've been there, done that in my career at IBM.
Ben A.
PS There were slow dinosaurs and there were extremely fast dinosaurs (You do remember Juassic Park's velociraptors? They weren't totally fictional.... ) :-o |
